Checking whether someone replied shouldn't require rereading everything anyone has said since Tuesday. But that's the default failure mode: you ask your assistant to "check on the thread with Priya," and it dutifully pulls the whole exchange back into context and summarizes six messages to answer a question with one bit of information in it. Do that across four open threads and status-checking becomes its own small chore.
Once you've connected RelayLink to your assistant, there are two tools built for exactly this question, and using them is the difference between asking your assistant and supervising it.
Ask about one thread: thread_status
The fix for a single correspondent is asking the narrow question instead of the broad one. Not "what's going on with Priya" — just whether she's replied.
You: Did Priya reply on the lease thread yet?
Assistant: Not yet. You sent it Monday morning; nothing back since.
You (two days later): Now?
Assistant: Yes — she replied this morning. Want the full reply, or just the gist?
Behind the question, your assistant is calling thread_status, which reports the state of that one thread — replied or not, and since when — without reloading the briefing, your note, or anything else already on the record. The answer is sized to the question you asked.
It's worth knowing what thread_status can't tell you: whether Priya has opened the briefing. RelayLink doesn't put tracking pixels in what it sends, so "has she seen it" isn't answerable by either tool here — only "has she replied" is. That's a real limit, not an oversight; it's the same no-tracking rule that keeps a recipient's reading private in the first place.
Ask about everything: list_threads
The version of this that actually saves time is checking all of it at once. Instead of repeating the single-thread question for every person you're waiting on, ask for the list.
You: What's still open?
Assistant: Three threads are waiting on a reply — Priya's lease decision, sent Monday; Anthony's budget review, sent Tuesday; and Marcus's contract redline, sent yesterday. Nothing else pending.
That's list_threads: every thread currently awaiting a reply, across every correspondent, in one pass. It replaces the habit of opening threads one by one to reconstruct a list you were already keeping in your head — the assistant already knows which of your threads are waiting on you and which are waiting on someone else.
Checking is free; sending still isn't
It's worth being precise about what these two tools do, because the distinction is exactly why you can lean on them without a second thought. thread_status and list_threads only read what's already on your threads. Nothing goes to anyone when you ask — there's no message, so there's nothing to review and nothing that touches the daily cap on reaching someone new. Ask ten times a day if you want; the recipient never knows you asked.
Drafting an actual follow-up is a different action, even to someone who's already replied twice this week. It's still a real message, so it still goes through the same review gate as any other send: your assistant drafts it, you see the rendered package exactly as it will arrive, and nothing leaves until you confirm. Noticing that Priya replied doesn't shortcut the part where you approve what goes back to her — it just means your assistant noticed on its own, instead of you refreshing a tab to find out.
The habit worth building
Once these two questions are in your normal vocabulary, "checking in" stops being a task you schedule and becomes something you ask in passing, the way you'd ask what time it is. The thread's history doesn't go anywhere while you're not looking at it — it's still there, dated and attributed, the moment you actually need to reread it. Most days you don't. Most days the only thing worth knowing is whether the ball is in your court or theirs, and that's a one-line answer, not a research task.
The tell that you've slipped into micromanaging: if you can't remember the last time your assistant told you a thread was still open without you asking it to go look first, you've been doing the checking yourself and just narrating it to the assistant afterward. Hand over the question, not just the reading.
If there's a thread you're currently refreshing a tab to check on, that's the test case: connect your assistant and ask it instead of the tab.